An adaptive protocol for efficient support of range queries in DHT-based systems

In recent years, Distributed Hash Tables (DHTs) have been proposed as a fundamental building block for large scale distributed applications. Important functionalities such as searching have been added to the DHT's basic lookup capability. However, supporting range queries efficiently remains a difficult problem. In this paper we describe an adaptive mechanism that relies on a logical tree data structure, the Range Search Tree (RST), to support range queries efficiently. Nodes in the RST automatically group registrations based on their values. Queries are decomposed into a small number of sub-queries for efficient resolution. The system dynamically optimizes itself to minimize the registration and query cost based on observed load. The system is fully distributed and avoids bottleneck problems encountered in traditional tree-based systems. Extensive simulation results validate the effectiveness of the system.